The network neutrality debate has been raging worldwide for around fifteen years now. Our goal in this paper is to model and discuss a quite recent option which could be seen as a trade-off between neutrality and differentiation operated by Internet service providers (ISPs), and satisfy both ends of the world: differentiation potentially chosen by end users. By using a model from the literature, we compare the outcomes of three scenarios: neutrality, non-neutrality with differentiation decided by ISPs, and non-neutrality decided by users. We illustrate that, depending on network parameters, letting end users decide may end up as a fair and viable solution, and that non-neutrality imposed by ISPs is not necessarily bad for all actors.